Method and device for image processing based on k-means clustering and dictionary learning

Abstract

The present invention provides a method and a device for image processing based on k-means clustering and dictionary learning. The method comprises the steps of: initializing image parameters to perform partitioning processing of a noise image, utilizing a k-means algorithm to perform clustering of a given image block, utilizing an MOD-SVD dictionary training algorithm to each clustering, obtaining a training clustering dictionary {D1, D2, D3, ..., Dk}, aggregating each partitioning dictionary {D1, D2, D3, ..., Dk} to form an over-complete dictionary D to make D={D1, D2, D3, ..., Dk}, obtaining each partitioning dictionary, forming the over-complete dictionary D, utilizing an OMP (Orthogonal Matching Pursuit) algorithm to solve a corresponding sparse coefficient, setting certain iterations, updating the dictionary and the sparse coefficient corresponding to learning, and reconstructing a denoising image. The PSNR (a peak signal to noise ratio) of an image denoising effect of algorithmprovided by the invention is compared with the PSNR of the image denoising effect of a known spare coding algorithm and an improved sparse coding (ISC) algorithm in the conditions of different noisevariances [Delta] and mixed noise densities d, and it is shown that the denoising effect in the invention can reach a better effect.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image processing method and device based on k-means clustering and dictionary learning. Background technique [0002] Entering the 21st century means entering the information age. With the development of science and technology, the form of information transmission is no longer limited to voice, but has developed into various multimedia forms including images, data, and text. When people receive external information, 80% come from visual information, and digital images are the most important source for people to obtain visual information. It has made great achievements, and has penetrated into many fields of people's daily life, study and work. The research content of digital image processing technology covers a wide range, including image denoising, image segmentation, image enhancement, image recognition, image restoration, image coding, multi-resolution processing,...

AI Technical Summary

Problems solved by technology

[0013] The embodiment of the present invention provides an image processing method and device based on k-means clustering and dictionary learning, which is used to solve the technical problems of insufficient traditional dictionary training and poor denoising effect

Images